arXiv subjects

Jaeseung Lee

Publications and source records attributed to Jaeseung Lee.

4 recordsLinked to original sources

arXiv2024

A Lightweight CNN-Transformer Model for Learning Traveling Salesman Problems

Several studies have attempted to solve traveling salesman problems (TSPs) using various deep learning techniques. Among them, Transformer-based models show state-of-the-art performance even for large-scale Traveling Salesman Problems (TSPs). However, they are based on fully-connected attention models and suffer from large computational complexity and GPU memory usage. Our work is the first CNN-Transformer model based on a CNN embedding layer and partial self-attention for TSP. Our CNN-Transformer model is able to better learn spatial features from input data using a CNN embedding layer compared with the standard Transformer-based models. It also removes considerable redundancy in fully-connected attention models using the proposed partial self-attention. Experimental results show that the proposed CNN embedding layer and partial self-attention are very effective in improving performance and computational complexity. The proposed model exhibits the best performance in real-world datasets and outperforms other existing state-of-the-art (SOTA) Transformer-based models in various aspects. Our code is publicly available at https://github.com/cm8908/CNN_Transformer3.

cs.LG↗

arXiv2023

Learning Geometric Combinatorial Optimization Problems using Self-attention and Domain Knowledge

Combinatorial optimization problems (COPs) are an important research topic in various fields. In recent times, there have been many attempts to solve COPs using deep learning-based approaches. We propose a novel neural network model that solves COPs involving geometry based on self-attention and a new attention mechanism. The proposed model is designed such that the model efficiently learns point-to-point relationships in COPs involving geometry using self-attention in the encoder. We propose efficient input and output sequence ordering methods that reduce ambiguities such that the model learns the sequences more regularly and effectively. Geometric COPs involve geometric requirements that need to be satisfied. In the decoder, a new masking scheme using domain knowledge is proposed to provide a high penalty when the geometric requirement of the problem is not satisfied. The proposed neural net is a flexible framework that can be applied to various COPs involving geometry. We conduct experiments to demonstrate the effectiveness of the proposed model for three COPs involving geometry: Delaunay triangulation, convex hull, and the planar Traveling Salesman problem. Our experimental results show that the proposed model exhibits competitive performance in finding approximate solutions for solving these problems.

cs.CG↗

arXiv2019

Strong solutions to the inhomogeneous Navier-Stokes-BGK system

In this paper, we are concerned with the local-in-time well-posedness of a fluid-kinetic model in which the BGK model with density dependent collision frequency is coupled with the inhomogeneous Navier-Stokes equation through drag forces. To the best knowledge of authors, this is the first result on the existence of local-in-time smooth solution for particle-fluid model with nonlinear inter-particle operator for which the existence of time can be prolonged as the size of initial data gets smaller.

math.AP↗

arXiv2019

A hydrodynamic model for synchronization phenomena

We present a new hydrodynamic model for synchronization phenomena which is a type of pressureless Euler system with nonlocal interaction forces. This system can be formally derived from the Kuramoto model with inertia, which is a classical model of interacting phase oscillators widely used to investigate synchronization phenomena, through a kinetic description under the mono-kinetic closure assumption. For the proposed system, we first establish local-in-time existence and uniqueness of classical solutions. For the case of identical natural frequencies, we provide synchronization estimates under suitable assumptions on the initial configurations. We also analyze critical thresholds leading to finite-time blow-up or global-in-time existence of classical solutions. In particular, our proposed model exhibits the finite-time blow-up phenomenon, which is not observed in the classical Kuramoto models, even with a smooth distribution function for natural frequencies. Finally, we numerically investigate synchronization, finite-time blow-up, phase transitions, and hysteresis phenomena.

math.AP↗